In the context of pressure ulcer prevention, this article deals with the problem of detecting and analysing micro-movements in the sacral area of bedridden patients on mattresses equipped with a network of pressure sensors. The study is based on a series of pressure measurements carried out on a cohort of patients lying on two types of mattress at the Nîmes university hospital (France). A spatio-temporal model considers first the local information of measurements from the array of sensors using an evidential filter in order to remove spatial uncertainties or measurement noise before micro-movement analysis. With a Detrended Fluctuation Analysis (DFA), the complexity level of the time series coming from the micro-movement model is finally estimated for different noise filters.